Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2022-06-06 Cooperative journals: 《计算机应用研究》
Abstract： Most of the existing traffic flow prediction algorithms only consider the prediction under normal conditions, but not the influence of weather attributes and surrounding geographical attributes on the prediction results, This paper proposed a combined prediction model (A-STIGCN) integrating external attributes . First, the external attributes are taken as the attributes of the sections in the road network, and the attributes and traffic characteristics of the sections are also put under modeling to obtain the enhanced feature vectors. Secondly, the method used graph wavelet transform and adaptive matrix to extract the local and global spatial feature information of the traffic flow respectively, with the help of the gating cycle unit (GRU) to extract the temporal information. Finally, the temporal dynamic variability of the attention mechanism is captured to predict the traffic flow. Shenzhen taxi trajectory data, corresponding weather data and POI data for prediction, the research results show that A-STIGCN combination model is better than the traditional linear model and variant model, compared with the ASTGCN model without introducing attention mechanism, MAE reduced about 0.131, accuracy improved 0.068, and the TGCN model without introducing external factors, MAPE. The reduction by about 0.637% and the improved accuracy by 0.079 provides better guidance for traffic management.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2019-01-28 Cooperative journals: 《计算机应用研究》
Abstract： Aimed at the problems of over- fitting of mapping function and insufficient convergence of loss function in convolution neural network image super- resolution algorithm, combined existing vision recognition algorithm and depth learning theory, this paper proposed an improvement on it. Firstly, the original SRCNN network increased layer number from 3 to 13 layers, and proposed a form of self- gated activation function Swish to replace the usual network model Sigmoi, ReLU and other activation functions, and fully utilized the advantages of Swish function to effectively avoid Over-fitting problems, better to learn and use the mapping relationship between low-resolution and high-resolution images to guide image reconstruction. Then introduced the Newton-Raphson method into the traditional network loss function, which further accelerates the convergence speed. Finally, experiments show that the improved network model can effectively improve the image definition, and improve the visual effect and objective parameter evaluation index.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-12-13 Cooperative journals: 《计算机应用研究》
Abstract： To solve the problems that the traditional direction-of-arrival(DOA) estimations have poor performance when processing coherent signals in the cases of low signal-to-noise ratio and non-uniform noise . This paper proposed a DOA method based on the modified matrix decomposition method, with the best covariance matrix by the convex optimization. The modified matrix decomposition method can deal with the extraction of the coherent sources, while, overcomes the aperture loss. Furthermore, the method reconstrtucted the noise free covariance matrix by the convex optimization. Finally, using minimization search to calculate DOA. The simulation results show that, comparing to the matrix decomposition (MD) algorithm, L1-norm singular vector decomposition(L1-SVD) algorithm and spatial smoothing based covariance rank minimization (SS-CRM) algorithm, the method suppress the non-uniform noise well and have good performance in low signal noise ratio (SNR) .
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-11-29 Cooperative journals: 《计算机应用研究》
Abstract： Recommender systems (RS) help users to find interesting information in plenty of data resources, and provide accurate personalized recommendation. While the recommendation algorithm based on information core can greatly reduce the time cost in the recommendation process. Aiming at the scalability problem in collaborative filtering recommendation algorithm, On the basis of the original information core extraction method based on frequency (frequency-based, FB) and ranking (rank-based, RB) , this paper proposes an improved extraction information core method IFB (IFrequency-based) and IRB(IRank-based) . When in search of the most similar neighbors, we proposed a concept : optimization set, and found the most similar neighbors for each user on this set. The experimental results showed that this method can get more accurate recommendation results, and reduce the mean average absolute error(MAE) effectively. At the same time, it has higher precision and recall, so it has better recommendation effect.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-20 Cooperative journals: 《计算机应用研究》
Abstract： In order to improve the accuracy of community detection for complex networks, this paper proposed an algorithm based on cuckoo search algorithm combining gene inheritance and greedy search (GGCSCA) to optimize modularity for community detection. Cuckoos walked randomly on ordered adjacent table and employed gene inheritance strategy, which aim to optimize population efficiently. The algorithm improved population quality quickly by greedy preference search of local modularity increment maximum for the purpose of getting good result of community partition. GGCSCA has been tested on both benchmark networks and some typical complex networks, and compared with some typical community detection algorithms. Experimental results show the effectiveness, accuracy and fast convergence of this algorithm for discovering community structure. It has strong capability of community identification and can detect the structure of community finely.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-18 Cooperative journals: 《计算机应用研究》
Abstract： With regards to the low precision and poor robustness of fuzzy C-means clustering algorithm (FCM) in the underdetermined mixing matrix estimation, this paper proposed an algorithm using genetic simulated annealing optimized FCM(GASA-FCM) based mixed clustering with hough transform to solve underdetermined mixing matrix estimation. It could combine the global search, high-precision advantages of simulated annealing algorithm(SA) and powerful search ability of space of genetic algorithm(GA) , FCM assigned the clustering center point obtained by genetic simulated annealing algorithm, which avoided the randomness of initial selection. The center of each kind of data obtained by clustering by hough transform is modified to improve the estimation accuracy of the mixing matrix. Experimental results show that the proposed algorithm significantly improves the stability of the algorithm and the accuracy of the mixing matrix estimation, and has certain validity and feasibility.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-05-02 Cooperative journals: 《计算机应用研究》
Abstract： In view of the fact that the existing traffic flow prediction methods ignore the efficient use of the traffic flow data characteristics and can not simulate more complex mathematical operations, a traffic flow prediction method is proposed to improve the deep belief network (DBN) . The proposed method combines deep belief network model and Softmax regression as prediction models, and uses the continuous restricted boltzmann machine (CRBM) to process the input eigenvectors and reduces the adaptive learning step (ALS) RBM trains the time needed to reconstruct the network model, and uses the improved deep belief network model to study the traffic flow characteristics. Softmax regression model is connected to the top of the network for traffic prediction. The experimental results show that in the actual traffic flow data prediction, the improved prediction accuracy and time complexity of the improved DBN model are better than the traditional prediction model.
Subjects: Computer Science >> Integration Theory of Computer Science submitted time 2018-04-24 Cooperative journals: 《计算机应用研究》
Abstract： With the development of communication technologies, heterogeneous networks with multiple access technologies have become the trend of future communication networks. With the increase of QoS requirements for user services and the increase of transmission bandwidth, existing network selection algorithms can no longer satisfy users' high quality communication needs. In view of the increasing shortage of spectrum resources in heterogeneous wireless networks, this paper proposed a two-level dynamic network selection scheme that involves both the user and the network. The scheme includes gray correlation analysis method and bipartite graph joint optimization matching algorithm. Through the joint decision of the user and the network, the algorithm optimizes the system throughput and equalizes the network load on the premise of effectively satisfying the quality of service requirements of mobile users. Simulation results show that compared with the traditional algorithm, the scheme greatly improves the utilization ratio of spectrum resources in heterogeneous network, reduces the handover probability of users switching between wireless networks, and realizes the rational allocation of users' needs and network resources.
Hits
Hits
Downloads
Comment